China’s New Standard Targets AI Customer Service Woes

China is rolling out its first national standard addressing one of the most frustrating experiences in modern digital life: getting stuck in an AI customer service loop with no way to reach a human. The new standard, which takes effect September 1, mandates clear transfer-to-human options and holds companies accountable for what their AI systems promise.

The Problem: Consumers Trapped in AI Loops

The frustration is widespread. A consumer from Xuzhou, Jiangsu, described her experience with an e-commerce platform’s AI customer service as “punching cotton” — after more than an hour of struggling through automated responses, she finally reached a human agent who solved the problem in minutes. On the Heimaotousu consumer platform, searches for “AI customer service” return over 4,700 complaints, many describing the same experience of being trapped in chatbot loops.

The scale of the issue is significant. According to China Consumers Association data, the organization received 985,928 consumer complaints in the first half of 2026, with after-sales service issues accounting for the highest proportion at 26.79 percent. AI customer service problems — including false promises, difficulty reaching human agents, and inaccurate generated content — have become a new hotspot.

A CCTV reporter’s test highlighted the problem vividly: saying “transfer to human” four times before the transfer process even began, and waiting nearly six minutes from dialing to reaching a human agent. The hashtag about the six-minute ordeal went viral on Chinese social media.

The Economics Behind the Problem

The root cause is largely economic. Human customer service agents cost approximately 3,000 RMB per month on average, while AI customer service subscriptions can be as low as 99 RMB per month — a roughly 30-fold cost difference. Large e-commerce companies can save over 1 million RMB annually by switching to AI, according to Economic Daily reporting.

Some companies have deliberately designed AI systems as an “invisible filter,” using complex menus and repeated loops to exhaust users’ patience and discourage complaints. “This is a mistaken approach to cost reduction,” said Li Ning, a professor at Tsinghua University’s School of Economics and Management. “On the surface, customer service costs are lower and manual service volume is reduced, but problems are not solved — they are merely postponed or transformed into repeated inquiries, complaints, and even returns and brand trust losses.”

The New Standard: GB/T 47746-2026

Issued by the State Administration for Market Regulation on May 25, the standard — formally titled “Customer Contact Service—Requirements for Collaboration between Live and Intelligent Customer Services” — takes effect September 1. It is China’s first national standard focused specifically on the collaboration mechanism between human and AI customer service, according to the official interpretation.

The standard requires that:

• Platforms must provide a clear “transfer to human service” menu option
• AI should handle standardized questions while human agents handle complex and safety-related issues
• Switching between AI and human must sync customer information without requiring repetition
• AI-generated responses must carry clear AI-generated labels and risk warnings
• Final confirmation for price, discounts, refunds, compensation, and contract changes must be done by human agents or standardized processes
• Automatic transfer to human agents is required when users explicitly refuse AI service, in emergencies involving personal or property safety, or when interaction thresholds are reached
• Special priority for elderly and disabled groups to access human agents

Legal Effect and Accountability

The standard is a recommended national standard rather than a directly enforceable law, but it serves as an important reference for regulators and courts in service disputes and quality assessments. “The progress of this new regulation is that for the first time, it clearly makes ‘transfer to human’ a standard feature rather than an optional one,” said Zhang Xiaorong, director of the Deep Technology Research Institute, as reported by Dazhong News.

Crucially, the China Consumers Association has clarified that AI customer service deployed by an operator constitutes part of the operator’s service delivery. Companies cannot refuse to honor commitments by claiming “AI responses don’t represent the company’s position.” This closes a loophole that has frustrated many consumers who received promises from AI chatbots only to have them denied by human agents.

Legal experts note that existing laws already require convenient complaint mechanisms. “If the layering exceeds reasonable limits — such as too many menu levels, deliberately hidden human entrances, no clear prompts, or AI unable to recognize ‘transfer to human’ keywords — it violates the legal obligation under Article 59 of the E-Commerce Law,” said Zhang Huan, a partner at Beijing Dacheng (Nanchang) Law Firm, as reported by China Youth News.

What This Means Going Forward

For consumers, the standard provides a new benchmark for evaluating customer service quality. “Consumers can save screen recordings, chat records, and failed transfer attempts as evidence, and file complaints with the company or through channels like 12315,” said Liao Huaixue, senior partner at Tahota Law Firm. “If actual losses occur, they can seek remedies through mediation, arbitration, or litigation.”

For companies, the standard signals a shift from treating AI as a cost-cutting tool to designing proper human-machine collaboration. As Li Ning put it, the standard “does not draw a static line of ‘this side for humans, that side for AI,’ but rather proposes a dynamic ‘human-machine collaboration’ logic.” Companies will need to redesign workflows, adjust performance metrics from “interception rate” to “resolution rate,” and ensure AI systems are properly governed and validated.

The standard also reflects China’s broader approach to AI governance — balancing technological innovation with consumer protection. As AI becomes increasingly embedded in everyday services, the question of who is accountable for AI’s actions is becoming central to regulatory frameworks worldwide. China’s new standard offers one answer: AI can assist, but humans must remain in control of the critical decisions.